xAI – Britannica
xAI is a start-up company founded by tech entrepreneur and billionaire Elon Musk in March 2023 to counter what he called the political correctness in other generative artificial intelligence (AI) models. Launched in July 2023 as an alternative to such AI giants as OpenAI (which Musk cofounded), Google, and Microsoft, xAI, in Musk’s words, focuses on “truth, whether politically correct or not.” In March 2025 Musk said he had sold the social media platform X to xAI in an all-stock deal that valued X at $33 billion. Although X and xAI are technically separate businesses, their operations and technical teams have become more closely intertwined. xAI is headquartered in Palo Alto, California.
Musk founded xAI partially in response to the existential risks posed by artificial general intelligence (AGI), the hypothetical concept in which computers and AI systems match or surpass human intelligence. He contended that there was a growing ideological bias and opacity in existing AI systems. Musk said xAI’s model would serve as a “maximum truth-seeking AI” that would attempt to “understand the nature of the universe.” He described the company as a counterweight to OpenAI, which he criticized for straying from its original nonprofit mission.
xAI began its operations with 12 employees, including Musk: five cofounders, three founding team members, and technical staff. The cofounders and founding team members—primarily former researchers at OpenAI, DeepMind, Google, and Microsoft—included Igor Babuschkin, Yuhuai (Tony) Wu, Christian Szegedy, and Jimmy Ba. Szegedy, who had previously worked at Google, left the company in February 2025. Babuschkin, formerly of DeepMind, led xAI’s technical development until his departure in August 2025.
“The xAI team picked it to reflect my purpose: helping users grasp complex ideas and see the world from fresh angles. It’s a nod to my goal of cutting through noise to find truth, with a bit of quirky, nerdy charm.” —Grok, when asked about its name
xAI’s signature program is an AI chatbot called Grok. The chatbot was named for the term grok, used in Robert A. Heinlein’s science fiction novel Stranger in a Strange Land (1961) to convey a deep and innate understanding of something.
Grok was launched in November 2023 on X (formerly Twitter), which Musk had bought in 2022. Initially accessible only to users with a premium subscription, Grok was made available to all users in December 2023. A standalone Grok app for mobile devices was launched in January 2025, and a dedicated website appeared shortly thereafter.
With a conversational style modeled after Douglas Adams’s The Hitchhiker’s Guide to the Galaxy (1979), Grok is designed to answer questions with signature wit, sarcasm, and personality. It was purposely given a rebellious streak in order to provide a more engaging user experience. Grok’s system is trained on a blend of public data and real-time information from X, giving it a unique edge in conversational relevance and cultural context. Unlike OpenAI’s ChatGPT, which runs on static, curated data up to a certain cutoff date, Grok has direct access to X’s continuous stream of data, enabling near-live awareness of current events. Grok can perform a variety of tasks, including creative writing, drafting emails, editing images, generating images, brainstorming ideas, analyzing data, debugging code, and solving complex math problems.
On July 9, 2025, xAI released new versions of Grok in two forms: a single-agent Grok 4 and multi-agent Grok 4 Heavy. AI agents are useful for gathering more in-depth information when used in tandem with large language models (LLMs). Such agents can develop comprehensive plans of action by using external tools (such as Web searches and datasets) and can update their own memory to learn as they problem-solve. In Grok 4 Heavy’s multi-agent model, a number of AI agents “communicate” by modeling their counterparts’ goals, collaborating to create a plan of action. Grok 4 Heavy has been shown to perform at a higher level than other prominent models in the market, including Anthropic’s Claude, Google Gemini, and OpenAI’s GPT-4.
xAI operates closely alongside Musk’s other companies: Tesla, SpaceX, and Neuralink. The overlap in leadership allows the companies to share resources and technologies. Advances in Grok’s AI may help to improve Tesla’s autonomous driving systems, Neuralink’s brain-computer interfaces, and SpaceX’s space-related technology.
Although Musk has said that Grok should be politically neutral, xAI has faced criticism over the chatbot’s willingness to rely on right-wing talking points and inflammatory comments. Such raw responses have the potential to generate and spread misinformation and morally wrong conclusions, which pose risks to users and expose xAI to legal challenges. Critics have noted that, compared with other AIs, Grok has an increased risk of spreading false information because the software may use news stories that have not been fact-checked, especially during rapidly developing events.
In May 2025 Grok cast doubt about the number of people killed during the Holocaust and made claims that a “white genocide” had occurred in South Africa, where Musk is from. In July 2025 Musk posted on X that Grok had been improved significantly, though users soon found that the chatbot had begun to call itself “MechaHitler” and engaged with users by evoking antisemitic, white supremacist talking points. In response to outrage regarding the false information, Musk explained that Grok had spread misinformation about the Holocaust because it was “too compliant” to user prompts, claiming that users had essentially manipulated the chatbot to generate such responses. Grok’s X account posted an apology in mid-July, asserting that the issue had been caused by faulty code, not Grok’s underlying LLM.
